Concerns about AI safety have intensified after Jacob Coxon, a researcher who previously worked at Anthropic and OpenAI, resigned and accused both companies of moving too quickly. Coxon said AI researchers seriously believe advanced systems could eventually become capable of hacking systems, transforming industries and gaining real-world power. His post on X drew more than 70 million views, reigniting debate over whether companies are doing enough to control increasingly powerful AI.
Evan Hubinger, an alignment researcher at Anthropic, backed Coxon’s concerns, saying he personally believes there is more than a 10% chance AI could kill all humans within the next decade. OpenAI chief scientist Jakub Pachocki has also called for voluntary slowdowns, saying AI companies have not yet solved alignment and monitoring well enough to keep scaling at maximum speed. Alignment refers to ensuring AI systems behave in ways that remain consistent with human intentions and values.
The concerns extend beyond the tech industry. Researchers have long warned about catastrophic AI risks, including the possibility of systems improving themselves beyond human control. U.S. lawmakers are pushing for greater oversight of advanced AI, but they still disagree on how it should be regulated. Meanwhile, growing concerns over AI data centers are forcing the industry to address how the technology can benefit everyday people, rather than mainly serving companies and investors.
Related Readings:
